How One Designer Led an AI Revolution at Pendo
A tutorial by Brian Greenbaum & Claire Vo. Featured in the OTF curated resource library.
The Spark
It started with frustration. A senior designer at Pendo had spent two weeks perfecting a dashboard redesign in Figma. The engineering implementation took another three weeks and, despite detailed specs, deviated from the design in dozens of small ways. Spacing was off. Animations were different. Edge cases were handled with generic patterns instead of the designed solutions.
This wasn't a communication failure — both teams were skilled and collaborative. It was a structural problem: translating static designs to interactive code always loses information.
The designer decided to try something different. They installed Cursor, loaded their Figma designs as reference, and spent a weekend building a working prototype. Monday morning, they presented working code instead of Figma screens. The response from engineering: 'This is exactly what we needed.'
The First Project
The first official project using the AI-assisted workflow was a redesign of Pendo's analytics dashboard. The designer:
1. Explored in Figma — rough concepts, layout exploration, design system updates
2. Prototyped in Cursor — built the dashboard using Pendo's actual React components and AI assistance
3. User tested with code — stakeholders and users interacted with a functional prototype
4. Handed off code, not mockups — engineering received a working prototype with ~60% production-ready code
The result: the project shipped 40% faster than comparable past projects. Design fidelity was dramatically higher. The engineering team spent time on business logic and performance instead of translating pixels.
The designer documented the entire process and shared it in Pendo's internal wiki. This documentation became the seed for company-wide adoption.
Spreading the Practice
How the practice spread from one designer to the entire design team.
Internal demo (Month 1)
The designer presented the dashboard project results at a company all-hands. The 40% speed improvement and design fidelity improvement got management attention immediately.
Peer learning sessions (Month 2-3)
Weekly 1-hour sessions where the pioneer designer helped colleagues try AI tools on their current projects. No pressure — just showing what's possible and helping with first steps.
Design team pilot (Month 4-5)
Three designers adopted AI tools for their next projects. Results: all three shipped faster with higher fidelity. The evidence was now impossible to ignore.
Organization-wide adoption (Month 6+)
Design leadership made AI tools available to all designers with optional training. ~70% of the design team adopted within 3 months.
Organizational Impact
Six months after the initiative:
Design-to-production time decreased 35% across the team. The biggest gain was eliminating the translation phase — designers hand off code, not mockups.
Design fidelity improved measurably. Internal surveys showed engineers rating design-code match at 8.5/10, up from 6/10 previously.
Designer job satisfaction increased. Designers reported feeling more ownership over the final product. The frustration of 'that's not what I designed' largely disappeared.
Engineering focus shifted. Frontend engineers spent more time on complex interactions, performance, and architecture rather than CSS adjustments and layout matching.
Hiring evolved. Pendo's design job postings now mention 'AI coding tool proficiency' as a plus. The role of 'design engineer' became an official career path.
The lesson: organizational change doesn't require top-down mandates. One passionate designer, one successful project, and a willingness to share knowledge can transform an entire team.
